Nvidia(US:NVDA) Xuan Gu Bao·2025-11-19 23:20Group 1 - Nvidia has decided to switch AI server memory chips from traditional DDR5 to low-power LPDDR chips, which are typically used in smartphones and tablets, to reduce energy costs [1] - This transition will position Nvidia as a significant LPDDR customer, comparable to major smartphone manufacturers, creating challenges for the supply chain to meet such demand [1] - Chip manufacturers are considering reallocating more factory capacity to LPDDR production to satisfy Nvidia's needs, potentially exacerbating supply tightness for other memory products [1] Group 2 - Memory prices are expected to rise by approximately 50% from current levels before the second quarter of 2026 due to critical chip shortages [1] - In October, various storage model prices accelerated, with month-on-month increases ranging from 40% to 100%, driven by the growing demand for AI servers [1] - The storage industry cycle is primarily driven by the explosion of storage demand in the AI era, with limited supply-side capacity, suggesting that the supply-demand gap may widen further into the first half of 2026, leading to sustained price increases [1] Group 3 - Domestic storage companies are expected to benefit from a dual drive of "price recovery + localization," with a robust recovery in demand, orders, and operating conditions anticipated in the fourth quarter [2] - Baive Storage has launched embedded storage products such as UFS3.1, LPDDR5/5X, and uMCP for AI smartphones, and has developed high-capacity LPDDR products of 12GB and 16GB [3] - Huicheng Co., Ltd. is providing LPDDR memory chip packaging services based on its PoP stacking packaging technology through its investment in Hefei Xinfeng Technology Co., Ltd. [3]